[Pellet-users] Reasoning over DIG and API

[sjtirtha]

Hi Evren,

I think, the problem is not on memory, but on the OWL model that I have. As
you said, DIG is less expressive than OWL. My model has a lot of "hasValue"
in the concept restriction. After deleting these "hasValue"s from my
model the integrated Pellet in Protege4.0 runs better. But it is still not
faster as reasonig over DIG interface. I've deleted also domain & range for
each properties, inverse properties, functional properties. Which other OWL
information are dropped in DIG ?

On my case I can not use DIG interface. Because the subsumption should also
subsumes some DataTypeRestriction (e.g. Weight <= 25 Kg). I've another
example that subsumes 10KgPackage to be a subclass of 25KgPackage using
DataTypeRestriction. Pellet infers it very well. But another example with
the box dimension is not inferred very well by Pellet. Example: Box A
(length:4, width:3, height:2) is subclass of Box B (length:6, width: 4,
height: 2). Does the implementation of OWL 1.1 support in Pellet already
mature ? How can I perform subsumption of DataTypeRestriction ?

Regards,

Steve


On 6/1/07, Evren Sirin <evren at clarkparsia.com> wrote:
>
> On 6/1/07 10:30 AM, sjtirtha wrote:
> > Hi,
> >
> > does Pellet distinct the mechanism of reasoning over DIG and API ?
> > I inferred my ontology over DIG interface and it goes very well and
> > fast. (Exactly I called "Classify taxonomy" & "Concept consistency" in
> > Protege)
> > Than I wrote a program used Pellet API and called some of reasoner
> > function( isConsistent(), isSubClassOf(), getAncestorClasses() ).
> > The program returned an error : OutOfMemoryError: Java heap space
> >
> > I thought, Pellet reasoner over API is faster. Did I do something wrong?
> DIG is always supposed to be slower than direct access because it has
> the overhead of HTTP communication and XML parsing/serialzation.
> However, DIG is known to be less expressive then OWL and cannot
> represent everything stated in an OWL ontology. Such  axioms (mostly
> things related to datatypes properties) will be dropped while Protege
> sends the ontology to the DIG server and you will see some warning
> messages on the Protege side. Sometimes those dropped axioms will make
> the ontology simpler to reason with.
>
> Reasoning algorithms are not deterministic so it is also normal to see
> fluctuations in the performance between different runs. Certain ordering
> of axioms make it easier to reason with the ontology. Depending on the
> order reasoner sees these axioms (which are affected by the internal
> hashtables) one run can be significantly faster than the other run.
> Unfortunately, there is no easy way for a reasoner to automatically
> figure out which ordering would be better than others.
>
> Cheers,
> Evren
> >
> > Steve
> > ------------------------------------------------------------------------
> >
> > _______________________________________________
> > Pellet-users mailing list
> > Pellet-users at lists.owldl.com
> > http://lists.owldl.com/mailman/listinfo/pellet-users
> > _______________________________________________
> >
> > Sponsored by Clark & Parsia, LLC http://clarkparsia.com/
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.owldl.com/pipermail/pellet-users/attachments/20070604/e4700173/attachment.htm